42 interview questions for developers (2026)

Easy

18 votes900 runs0 comment

What is the missing word in the following sentence: Overfitting is the production of a model that corresponds too closely to the training data set and may therefore fail to fit other data sets and so fail to perform any reliable forecast.

Data Scientist

overfitting

Martinat Mar 14, 2022

Easy

16 votes875 runs0 comment

What is the relation between Artificial intelligence (AI), machine learning and deep learning?

Data Scientist

machine learning

deep learning

Martinat Sep 19, 2022

Easy

8 votes881 runs0 comment

Given a neural network with 4 inputs, how many connections could exist between the input layer and the first hidden layer?

Data Scientist

neural network

connections

Martinat Apr 4, 2022

Easy

6 votes810 runs0 comment

Which of the following processes are machine learning ones but not particularly related to deep learning?

Data Scientist

machine learning

random forest

Martinat Mar 14, 2022

Easy

4 votes786 runs0 comment

Cross validation is a technique used for evaluating DL (deep learning) models (an ML models in general). The method consists in training several models on subsets of the input data and evaluating them on the complementary subset of this same data. It aims at avoiding overfitting.

Data Scientist

overfitting

validation

Martinat Apr 4, 2022

Easy

2 votes792 runs

2 comments

In order to get good results when implementing deep learning, one needs to prepare his data before using it. What action CAN be done during this preparation?

Data Scientist

data

preprocessing

Martinat Apr 4, 2022

Easy

1 vote847 runs0 comment

Which neural network type is the best for image processing?

Data Scientist

neural networks

image processing

Martinat Sep 19, 2022

Easy

1 vote829 runs0 comment

How to define unsupervised learning? In unsupervised learning ...

Data Scientist

machine learning

unsupervised learning

Martinat Mar 14, 2022

Easy

1 vote824 runs

2 comments

In deep learning, the validation set is used to provide frequent and unbiased evaluations of the model’s fit on the training set while tuning its hyperparameters/parameters: in other words,the model is found and then tested on the validation set before to be improved once again.

Data Scientist

hyperparameters

validation

Martinat Mar 14, 2022

Easy

1 vote819 runs1 comment

What type of learning are the following task included?

Data Scientist

supervised learning

classification

Martinat Mar 14, 2022

Easy

1 vote795 runs0 comment

What is the aim of data normalization? Its aim is to transform data sets so that all of its values are in a given range (ex: 0 to 1) while keeping the model's statistical properties.

Data Scientist

data

normalization

Martinat Mar 14, 2022

Medium

1 vote785 runs0 comment

What is/are the purpose(s) of a neuron’s activation function?

Data Scientist

activation function

non-linearity

Martinat Apr 4, 2022

Easy

1 vote784 runs0 comment

What are the parameters that are set before the learning process begins in deep learning?

Data Scientist

parameters

neural network

Martinat Mar 14, 2022

Easy

1 vote519 runs0 comment

What is a sigmoid function?

Data Scientist

activation function

sigmoid function

Martinat Mar 14, 2022

Easy

1 vote238 runs0 comment

How to define supervised learning? In supervised learning ...

Data Scientist

supervised learning

machine learning

Martinat Mar 14, 2022

Easy

1 vote236 runs0 comment

Which neural network type is the best for image processing?

Data Scientist

neural networks

image processing

Martinat May 6, 2022

Medium

1 vote155 runs0 comment

The learning rate is the variable which determines the gradient descent algorithm steps’ size. It is used when studying the cost function. How is this value chosen? What are the risks?

Data Scientist

gradient descent

cost function

Martinat Apr 4, 2022

Medium

1 vote79 runs1 comment

How many types of gradient descent algorithms are there?

Data Scientist

gradient descent

optimization

Martinat Mar 14, 2022

Medium

1 vote74 runs0 comment

Which of these statements are false regarding the “loss function”?

Data Scientist

loss function

machine learning

Martinat Apr 4, 2022

Medium

1 vote71 runs0 comment

Calculate the output of a neuron with 3 inputs and 3 weights.

Data Scientist

activation function

linear function

Martinat Mar 14, 2022

42 interview questionsfor developers (2026)

Test your developer skills before an interview with 42 corrected questions on 80+ topics. Beginner, Intermediate or Expert level.

What is the missing word in the following sentence: Overfitting is the production of a model that corresponds too closely to the training data set and may therefore fail to fit other data sets and so fail to perform any reliable forecast.

What is the relation between Artificial intelligence (AI), machine learning and deep learning?

Given a neural network with 4 inputs, how many connections could exist between the input layer and the first hidden layer?

Which of the following processes are machine learning ones but not particularly related to deep learning?

Cross validation is a technique used for evaluating DL (deep learning) models (an ML models in general). The method consists in training several models on subsets of the input data and evaluating them on the complementary subset of this same data. It aims at avoiding overfitting.

In order to get good results when implementing deep learning, one needs to prepare his data before using it. What action CAN be done during this preparation?

Which neural network type is the best for image processing?

How to define unsupervised learning? In unsupervised learning ...

In deep learning, the validation set is used to provide frequent and unbiased evaluations of the model’s fit on the training set while tuning its hyperparameters/parameters: in other words,the model is found and then tested on the validation set before to be improved once again.

What type of learning are the following task included?

What is the aim of data normalization? Its aim is to transform data sets so that all of its values are in a given range (ex: 0 to 1) while keeping the model's statistical properties.

What is/are the purpose(s) of a neuron’s activation function?

What are the parameters that are set before the learning process begins in deep learning?

What is a sigmoid function?

How to define supervised learning? In supervised learning ...

Which neural network type is the best for image processing?

The learning rate is the variable which determines the gradient descent algorithm steps’ size. It is used when studying the cost function. How is this value chosen? What are the risks?

How many types of gradient descent algorithms are there?

Which of these statements are false regarding the “loss function”?

Calculate the output of a neuron with 3 inputs and 3 weights.

42 interview questionsfor developers (2026)

Test your developer skills before an interview with 42 corrected questions on 80+ topics. Beginner, Intermediate or Expert level.

What is the missing word in the following sentence: Overfitting is the production of a model that corresponds too closely to the training data set and may therefore fail to fit other data sets and so fail to perform any reliable forecast.

What is the relation between Artificial intelligence (AI), machine learning and deep learning?

Given a neural network with 4 inputs, how many connections could exist between the input layer and the first hidden layer?

Which of the following processes are machine learning ones but not particularly related to deep learning?

Cross validation is a technique used for evaluating DL (deep learning) models (an ML models in general). The method consists in training several models on subsets of the input data and evaluating them on the complementary subset of this same data. It aims at avoiding overfitting.

In order to get good results when implementing deep learning, one needs to prepare his data before using it. What action **CAN** be done during this preparation?

Which neural network type is the best for image processing?

How to define unsupervised learning? In unsupervised learning ...

In deep learning, the validation set is used to provide frequent and unbiased evaluations of the model’s fit on the training set while tuning its hyperparameters/parameters: in other words,the model is found and then tested on the validation set before to be improved once again.

What type of learning are the following task included?

What is the aim of data normalization? Its aim is to transform data sets so that all of its values are in a given range (ex: 0 to 1) while keeping the model's statistical properties.

What is/are the purpose(s) of a neuron’s activation function?

What are the parameters that are set before the learning process begins in deep learning?

What is a sigmoid function?

How to define supervised learning? In supervised learning ...

Which neural network type is the best for image processing?

The learning rate is the variable which determines the gradient descent algorithm steps’ size. It is used when studying the cost function. How is this value chosen? What are the risks?

How many types of gradient descent algorithms are there?

Which of these statements are *false* regarding the “loss function”?

Calculate the output of a neuron with 3 inputs and 3 weights.

In order to get good results when implementing deep learning, one needs to prepare his data before using it. What action CAN be done during this preparation?

Which of these statements are false regarding the “loss function”?